KSpagnol@massey.ac.nz (Ken Spagnolo) (10/26/90)
(Sorry if this has been discussed recently.) We're dealing with Ultrix 4.0 here, with the Sep. 7 patch of c-news and nntp-1.5.10 (though that shouldn't matter). When my disk fills to the point where nntp stops accepting new batches, expire also stops expiring. This has happened twice. The first time, it started up again on its own a few days later and I don't remember any details, but the second time it went for a whole week not expiring anything. Throughout the week, I received expire's verbose output thru the mail as so: expire problems: 22820 kept, 0 expired 0 residual lines 0 links archived, 0 junked, 0 missing except that the residual line count went from about 11000 to 0. First question: What are residual lines? This made me believe that expire didn't have any disk space problems. I archive news to a different partition and have modified spacefor so that it has no problems letting expire get on with it (enough room for the new history files and expired articles, etc.). I also tried the -g option to report funny dates and rebuilt the history file from scratch. But nothing worked until I removed by hand several Mb of articles (no one reads news from this machine yet, so I can play around a bit). After another history rebuild, expire started working again. So it seems the partion was too full even though spacefor was happy? (Unfortunately, /usr/spool/news is not a separate partition on my system, in case you're wondering.) I removed 25 Mb before trying again, so I don't really know what the magic number would be, if it really is a space problem. What other factors are invloved with expiring articles that I don't know about? When expiring, is spacefor called to report on available space in any other areas than control and archive? I wouldn't think expire would go all the way thru and look at each article (or at least count them) if space was the problem. My usual expire command reads as follows: doexpire -v -a /expiredir I appreciate any tips and will post the winning solution for others. -- Ken Spagnolo - Systems Programmer, Postmaster, Usenet Administrator, etc. Computer Centre, Massey University, Palmerston North, New Zealand K.Spagnolo@massey.ac.nz Phone: +64-63-69099 x8587 New Zealand = GMT+12
henry@zoo.toronto.edu (Henry Spencer) (10/27/90)
In article <1104@massey.ac.nz> K.Spagnolo@massey.ac.nz (Ken Spagnolo) writes: >When my disk fills to the point where nntp stops accepting >new batches, expire also stops expiring... The first question to ask is "how is NEWSMASTER defined, and is mail to that address getting read?". Doexpire has a couple of fallback modes, each of which is accompanied by mail to NEWSMASTER reporting on the situation. If space is too short for archiving, expire gets invoked with -h; if space is too short for a history-file rebuild, expire gets invoked with -r. `expire -h -r' might explain what you're seing. >... First question: What are residual lines? The terminology is a little obscure, and needs fixing. Residual lines are history lines retained after the corresponding articles have expired. >it seems the partion was too full even though spacefor was >happy? ... I think you need to take a good hard look at those local changes to spacefor; sounds like they are not doing what you think they are. >... When expiring, is >spacefor called to report on available space in any other >areas than control and archive? No. > I wouldn't think expire >would go all the way thru and look at each article (or at >least count them) if space was the problem... Expire counts lines in the history file regardless of whether it can do anything about them. If you archive everything and expire is invoked with -h and -r, there really isn't anything it can do. -- The type syntax for C is essentially | Henry Spencer at U of Toronto Zoology unparsable. --Rob Pike | henry@zoo.toronto.edu utzoo!henry
K.Spagnolo@massey.ac.nz (Ken Spagnolo) (10/29/90)
(If anyone saw my previous response before I cancelled it, please ignore. It was bogus. I shouldn't rely on my memory so much :-) In article <1990Oct26.173652.805@zoo.toronto.edu> you write: >I think you need to take a good hard look at those local changes to >spacefor; sounds like they are not doing what you think they are. Thanx for your response. I did just that and came out with a bit of egg on my face. I neglected to change the partition spacefor df's for archive from $NEWSARTS, which certainly wouldn't have enough space on it to archive if it was filled up to start with. I archive to a different partition. There is a problem, though, with doexpire. Really, its a problem with sh. The reason I didn't realize early on that I'd goofed, was because the mail I got informing me the -h option was being used on expire was empty. This could be due to a bug I've heard about in some Bourne shells that won't let you pipe the result of ( ... ) anywhere. I changed doexpire to mail $NEWSMASTER <<ENDOFMAIL ... ENDOFMAIL and it now works. This problem exists with sh5 as well. You can do it interactively, but not from another shell. Oh well, more Ultrix work arounds. -- Ken Spagnolo - Systems Programmer, Postmaster, Usenet Administrator, etc. Computer Centre, Massey University, Palmerston North, New Zealand K.Spagnolo@massey.ac.nz Phone: +64-63-69099 x8587 New Zealand = GMT+12
henry@zoo.toronto.edu (Henry Spencer) (10/30/90)
In article <1990Oct29.045714.8306@massey.ac.nz> K.Spagnolo@massey.ac.nz (Ken Spagnolo) writes: >... the mail I got informing me the -h option was being used >on expire was empty. This could be due to a bug I've heard about >in some Bourne shells that won't let you pipe the result of ( ... ) >anywhere... Curious; I'd never heard of that one before, and it would break an awful lot of shell scripts. It's not a generic Ultrix problem, since the code from expire works fine on Ultrix on our 2100. -- "I don't *want* to be normal!" | Henry Spencer at U of Toronto Zoology "Not to worry." | henry@zoo.toronto.edu utzoo!henry
K.Spagnolo@massey.ac.nz (Ken Spagnolo) (10/31/90)
In article <1990Oct29.174750.19269@zoo.toronto.edu> henry@zoo.toronto.edu (Henry Spencer) writes: >In article <1990Oct29.045714.8306@massey.ac.nz> K.Spagnolo@massey.ac.nz (Ken Spagnolo) writes: >>. This could be due to a bug I've heard about >>in some Bourne shells that won't let you pipe the result of ( ... ) >>anywhere... > >Curious; I'd never heard of that one before, and it would break an awful >lot of shell scripts. It's not a generic Ultrix problem, since the code >from expire works fine on Ultrix on our 2100. Its even stranger. The problem only occurs when what is being piped to mail is over a certain length, as is the case with the -h and -r warnings. This is rather fortunate, since you use this method throughout c-news. All of the one liners, etc., work just fine. I experimented by arbitrarily removing some of the lines in question and, under a certain point, the body of the message started appearing the delivered mail. I didn't try to find out if it was the total length of text or the number of lines echoed that decided success. Guess I'll talk to DEC. -- Ken Spagnolo - Systems Programmer, Postmaster, Usenet Administrator, etc. Computer Centre, Massey University, Palmerston North, New Zealand K.Spagnolo@massey.ac.nz Phone: +64-63-69099 x8587 New Zealand = GMT+12